Fabric: An Agentic AI Workspace for Natural-Language Data Analysis and Multi-Tool Task Execution

Agentic artificial intelligence systems combine language-model interaction with tools that perform operations beyond text generation. In data analytics, an integrated workspace must support dataset handling, analytical computation, visualization, and controlled tool execution while making the limits of each capability clear. This paper presents Fabric, a software project intended to bring conversational model access and data-oriented workflows into a common workspace. The paper documents its architecture and a preliminary local evaluation on Windows 11 using Node.js v24.15.0. JavaScript syntax validation passed, and a targeted evaluation suite reported seven passing tests. A synthetic local benchmark processed 10,000 rows and verified a sum of 505,000. The broader test suite reported 415 passing tests and four failures, including a Windows filesystem-permission verification failure and spreadsheet-processing failures. These observations support only narrow claims about selected components. They do not establish end-to-end model accuracy, secure execution isolation, production readiness, or performance under real-world workloads. The paper reports the methodology, results, limitations, and further validation required for a rigorous assessment of Fabric as an AI-assisted data-analysis workspace.